Short-term aggregated load and distributed generation forecast using fuzzy grouping approach

The increasing shares of renewable energy sources at low voltage distribution nodes are the cause of increased operational uncertainty. This uncertainty must also be taken into account during operational planning for the short term period, i.e. up to five days ahead. Therefore the system operators must take into account how low-voltage load as well as generation change at each node in the near future. This requires appropriate short-term forecast load and distributed generation (DG) methods. In the past, standard profiles for nodes were used to approximately forecast the expected load in the absence of smart metering and historical time-series storage. With increased number of smart meters in the grid, it is possible to use more complex methods to more accurately forecast system load and generation and therefore ensure secure system operation. In this paper, we present the forecasting approach used for load and distributed generation nodes equipped with smart meters. The presented approach was designed at SIEMENS as one part of the Austrian nationally founded research project ProAktivNetz. The forecasts represent aggregated load and distributed generation nodal injection values. The approach was tested for a small node with 100 households and a 34 kWp photovoltaic power plant. The results show that the grouped forecast results using the fuzzy weighting procedure give better forecasting results than individual methods and significantly better results than using standard profiles for the nodes.